变电站高压电气设备局部放电检测技术分析

Analysis of partial discharge detection technology for high voltage electrical equipment in substation

为了准确检测变电站高压电器设备的局部放电故障,使用多轴平行摄影系统同时采集基于可见光、紫外线、红外线等的多频谱视频数据,使用多咪头采集现场的噪声数据,并采用三点定位法分别实现故障位置定位。在机器学习架构设计中,使用6个独立的归一化神经网络模块对6个探头的数据进行归一化处理,使用1个神经网络核心模块进行二值化分析,最终得到分析结果。仿真分析表明,该系统可对1×108 pF规模的放电现象实现敏感性100%的捕捉和判断;可以在较小的工程量基础上,实现对高压设备微小隐性故障的早期发现和早期处理,可以有效提升变电站的状态检修水平。

In order to detect the partial discharge fault of high-voltage electrical equipment in substation,it applies the technology collects video data based on visible light,ultraviolet light,infrared and other multi spectrum through multi axis parallel photography system,and combines multi microphone recording technology to collect stereo noise data.The video and noise data are used to capture the discharge process synchronously and locate the fault location respectively.In the design of machine learning architecture,six independent normalized neural network modules are used to normalize the data of six probes,and one neural network core module is used for binary analysis,and finally the analysis results are obtained.The simulation results show that the system can capture and judge the discharge phenomenon of 1×108 pF scale with 100%sensitivity.Finally,it is proved that the system can detect and deal with the hidden faults of high-voltage equipment in the early stage on the basis of a small amount of work,effectively improve the level of condition based maintenance of substation.